JSON-LD is a way of adding structured data to a web page as a block of JSON inside a script tag, using the schema.org vocabulary to state facts such as who published the page, who wrote it and what the page is about. Google recommends JSON-LD over other structured data formats. For AI search, its main value is removing ambiguity: it tells machines exactly which organisation, person and topic a page refers to.
Where schema.org comes from
Schema.org was launched in 2011 by Google, Microsoft (Bing) and Yahoo, and Yandex joined soon after. It is a shared vocabulary of types (Organization, Person, Article, Product, Service and hundreds more) and properties (name, address, author, datePublished). JSON-LD is one of three formats for writing it, alongside Microdata and RDFa, and it is the easiest to maintain because it sits apart from the visible HTML.
The parts that matter
| Keyword | What it does |
|---|---|
| @context | Declares the vocabulary, almost always https://schema.org |
| @type | The kind of thing being described, such as Organization or Article |
| @id | A stable identifier (usually a URL with a fragment) that other nodes can point to |
| @graph | A list of several nodes in one block, so they can reference each other |
| sameAs | Links to the same entity on other sites, such as LinkedIn or Wikidata |
A connected graph
The common mistake is to add schema piece by piece: an Organization block from one plugin, an Article block from another, author details typed as plain text. The result is several disconnected descriptions. A connected graph uses @id so that each node refers to the others:
{
"@context": "https://schema.org",
"@graph": [
{
"@type": "Organization",
"@id": "https://www.example.com/#organization",
"name": "Example Ltd",
"url": "https://www.example.com/",
"sameAs": ["https://www.linkedin.com/company/example"],
"founder": { "@id": "https://www.example.com/#jane-doe" }
},
{
"@type": "Person",
"@id": "https://www.example.com/#jane-doe",
"name": "Jane Doe",
"worksFor": { "@id": "https://www.example.com/#organization" },
"sameAs": ["https://www.linkedin.com/in/janedoe"]
},
{
"@type": "Article",
"headline": "How we cut page load time by half",
"author": { "@id": "https://www.example.com/#jane-doe" },
"publisher": { "@id": "https://www.example.com/#organization" },
"datePublished": "2026-03-10",
"dateModified": "2026-06-02"
}
]
}This site uses the same pattern: the Organization node lists both co-founders by @id, and each Person node links back to the Organization and to the founder’s LinkedIn profile.
Rich results versus understanding
Google only shows special search features (rich results) for a limited list of schema types, and it has narrowed that list over time. FAQ rich results have been limited to well-known government and health sites since August 2023, and HowTo rich results are no longer shown. Many schema types never produce a visible feature.
That does not make them useless. Google’s documentation says structured data helps it understand page content, and a clear graph reduces the chance that your company is confused with another of a similar name. How much weight AI assistants give to JSON-LD is not documented by any provider, so we treat it as a clarity measure that supports, but does not replace, clear visible content.
Common problems we find
- Two Organization nodes with different names or logos, output by a theme and an SEO plugin at the same time
- Authors given as a plain name string instead of a Person node
- Schema describing content that is not visible on the page, which goes against Google’s guidelines
- dateModified updated automatically on every save, including trivial edits
- FAQPage schema added for rich results that Google no longer shows for most sites
How to check your schema
- Google’s Rich Results Test shows which rich results a page qualifies for and any errors.
- The Schema Markup Validator at validator.schema.org checks all schema.org types, including those Google does not use for rich results.
- View the page source and search for “application/ld+json” to count how many separate blocks exist. More than one Organization node is a warning sign.
Frequently asked questions
Does schema improve rankings?
Google has said structured data is not a general ranking factor. It can make a page eligible for rich results, which can change click-through rates, and it helps search engines understand content.
Do I need an SEO plugin for schema?
No. Plugins such as Yoast and Rank Math generate a graph automatically, which suits most sites. A custom theme can output its own graph, as this one does. What matters is that only one source outputs the core Organization and WebSite nodes.